Problem/misunderstanding with ObjectCache, multiple databases and ObjectModification, 1.0.4

Hi,

I have some problems understanding (and using) the concept of the cache when using multiple databases and determining the object modification level at the store method.

Assume we have two databases A and B with one being the default and one being the backup database. Both connections are configured using the cache:
       <object-cache class="org.apache.ojb.broker.cache.ObjectCacheDefaultImpl">
         <attribute attribute-name="timeout" attribute-value="-1"/>
       </object-cache>

Given the following code sequence with empty databases, I have some problems:

AnObject someObject = new AnObject();
[...]
PersistenceBroker defaultBroker = PersistenceBrokerFactory.defaultPersistenceBroker();
defaultbroker.store(someObject); // resulting in an INSERT to the default DB
[...]
defaultBroker.store(someObject); // resulting in an UPDATE to the default DB

PersistenceBroker backupBroker = PersistenceBrokerFactory.createPersistenceBroker(new PBKey("backup"));
backupBroker.store(someObject); // resulting in an UPDATE to the backup DB, which fails as there is no such row in the DB


As far as I understand the code from PersistanceBrokerImpl, it should query the cache and/or the DB:
        /*
        if PK values are set, lookup cache or db to see whether object
        needs insert or update
        */
        if (!insert)
        {
            insert = objectCache.lookup(oid) == null
                && !serviceBrokerHelper().doesExist(cld, oid, obj);
        }
        store(obj, oid, cld, insert);

The insert should be false, as the PK fields are not null, and the objectCache for the backupBroker should give null.
I traced with P6Spy, but I can't detect a SELECT to determine the existence of the row, so I assume the objectCache.lookup(oid) returns true.

Any hints?

Bernd